Advertisement

The social footprint—a practical approach to comprehensive and consistent social LCA

  • Bo P. Weidema
SOCIAL LCA IN PROGRESS

Abstract

Purpose

The practicality of social footprinting is currently hampered by an excessive data requirement, a lack of focus on materiality of the impacts, and a lack of understanding of the main impact pathways (cause-effect relationships) for social and economic impacts. We propose a “streamlined” method to overcome these barriers without loss of comprehensiveness.

Methods

The method combines a top-down approach using input-output data to focus the data collection effort on processes with high value added or a high number of work hours with a streamlined impact assessment that limits the inventory data requirement and the need for detailed impact pathway descriptions, by focusing on the macro-scale impacts of income redistribution and productivity impacts of missing governance, both of which can be classified as nonproduction-specific impacts, i.e., unrelated to enterprise-specific actions and choice of technology, and therefore quantifiable from national statistics without need to access detailed technology- or enterprise-specific data. The method is open for further refinement and detail in areas of specific interest for a particular product or project.

Results and discussion

We show that nonproduction-specific impacts constitute the vast majority of social and economic impacts and how important income inequality is for the impact assessment. We apply a novel way of combining impacts on productivity and impacts on human well-being and show that inequality implies that an intervention that changes the amount of QALY (quality-adjusted life years) for a population group will always give a larger change in well-being than an intervention of the same monetary value that only affects the level of consumption of the same population group. Throughout the article, we apply and illustrate the method with an example from the clothing industry.

Conclusions

The presented method allows comprehensive assessments of social footprints of products at much lower efforts than seen so far. The potential credits for positive action is by far the largest in countries with missing governance, thus providing a compelling argument for placing activities in countries with missing governance, provided that it allows the enterprise to follow an active strategy to create shared value.

Keywords

Governance Income inequality Income redistribution Input-output data Non-production-specific impacts Productivity Quality-adjusted time Shared value 

1 Introduction

We propose a practical, yet comprehensive approach to accounting and assessment of social footprint. We use here the term “social” as it is used in welfare economics, to signify an accounting that encompasses the entire societal economy, as in “social costs”, combining private costs and externalities. The “social footprint” is thus to be understood as the result of a complete “life cycle sustainability assessment” (LCSA) as suggested by Klöpffer (2008). Thus, the social footprint could also be called a “sustainability footprint”.

Existing LCSA methods are characterised by an excessive data requirement (Andrews et al. 2009; Ciroth and Franze 2011), requiring the choice between and/or calculations on a long list of social impact categories, for example 157 indicators for 33 impact categories in the PROSA method (Grießhammer et al. 2007), or 150 indicators for 22 impact categories in the social hotspot database (Benoit-Norris et al. 2012). With the exception for Weidema’s method (Weidema 2006a,b), there is no social and economic impact pathway (cause-effect relationship) modelling implemented in the existing methods (Falque et al. 2014; Reitinger et al. 2011), which means that there is no procedure for distinguishing important from less important among the many indicators. Furthermore, existing LCSA methods have an (unwarranted, as we shall show) focus on enterprise-specific data, which has resulted in practically no LCSA studies being able to reach beyond a very rudimentary product system (Jørgensen et al. 2008, 2010a; Macombe et al. 2013). The most complete product systems currently available are those of the hotspot database (http://socialhotspot.org/), which contains data on work hours, value added and semi-quantitative ‘risk’ data for 227 countries and 57 economic sectors. However, the risk indicators are intransparent because the original data are not included, and some serious errors in the value added data have not yet been fixed.

In conclusion, the practicality of social footprinting is currently hampered by an excessive data requirement, a lack of focus on materiality of the impacts, and a lack of understanding of the main impact pathways (cause-effect relationships) for social and economic impacts.

Our proposal to overcome these barriers is a practical “streamlined” method for accounting and impact assessment, without loss of comprehensiveness. The method combines a top-down approach to focus the data collection effort on impact categories and inventory data of high importance with a streamlined impact assessment that limits the inventory data requirement and the need for detailed impact pathway descriptions, by focusing on the macro-scale impacts.

Our impact assessment method applies a novel way of combining impacts on productivity and impacts on human well-being, taking into account income inequality, and shows how important income inequality is for the assessment of the other impacts.

The method is open for further refinement and detail in areas of specific interest for a particular product or project.

This article first describes how input-output data can be used to focus the data collection effort on processes with high value added or a high number of work hours, since all social and economic impacts are directly or indirectly linked to one of these two variables (Section 2 “Focusing on the important impact categories and activities”). Section 3.1 describes the impact of income redistribution, while Section 3.2 describes the productivity impact of missing governance, both of which can be classified as nonproduction-specific impacts, i.e., unrelated to enterprise-specific actions and choice of technology, and therefore quantifiable from national statistics without need to access detailed technology- or enterprise-specific data. That such nonproduction-specific impacts constitute the vast majority of social and economic impacts is demonstrated in Section 3.3 “The relative importance of nonproduction-specific impacts”. The importance of inequality and marginal utility, which is highlighted already in Section 3, is further emphasised in Section 4. “The importance of inequality for valuing changes in quality-adjusted time” where it is shown that inequality and marginal utility implies that an intervention that changes the amount of QALYs for a population group will always give a larger change in well-being than an intervention of the same monetary value that only affects the level of consumption of the same population group. Finally, Section 5 addresses the potential credits for positive action, which is by far the largest in countries with missing governance, thus providing a compelling argument for placing activities in countries with missing governance, provided that it allows the enterprise to follow an active strategy to create shared value. Throughout the article, we apply and illustrate the method with an example from the clothing industry.

2 Focusing on the important impact categories and activities

While a social footprint should be based on a complete inventory and a coverage of impact categories that in its outset should be both exhaustive and nonoverlapping, this does not mean that all impacts need to be recorded in equal detail.

By applying a top-down approach, in which inventory data for an exhaustive set of impact categories are estimated at the global level (Weidema 2006a), and assessed by expressing the impact in comparable (monetary) units, as described in the following sections on impact assessment, it is possible to focus on the relatively few impacts that dominate in global importance, rather than spending time and efforts on obtaining data for a large number of overlapping and largely irrelevant indicators (Jørgensen et al. 2010b; Falque et al. 2014).

Since all social and economic impacts are directly or indirectly linked to either the working time or the value added of the productive activities,1 the collection of specific data can be streamlined by focussing only on processes with high value added or high number of work hours. Before high quality data are available on these activities, it is meaningless to collect better quality data on activities that contribute less to the overall impact.

As an example, Table 1 shows how high value added activities and activities with a high number of work hours can be identified for a specific life cycle, here an average clothing item purchased in France, by the use of a multi-regional input-output database, here Exiobase v.1 (Tukker et al. 2013). The relatively high number of low-paid work hours for manufacturing of apparel in Asia is explained in Kabeer and Mahmud (2004) by “the limpness of materials used in clothing has made mechanisation in the labour-intensive assembly stage extremely difficult and, as long as there are plentiful supplies of cheap labour available, uneconomical.”
Table 1

Example: Activities with high value added and high number of work-hours in the upstream life cycle of an average clothing item bought in France. Data calculated from Exiobase v.1

Activity

Value added

Work hours

Animal products, China

<1 %

3 %

Apparel and textile manufacture, rest of Europe

7 %

5 %

Apparel and textile manufacture, France

13 %

3 %

Apparel and textile manufacture, India

1 %

12 %

Apparel and textile manufacture, rest of Asia

6 %

12 %

Chemicals manufacture, Asia

1 %

4 %

Cultivation of plant-based fibres, India

<1 %

0.4 %

Electricity, Asia

<1 %

2 %

Extraction of crude petroleum, Asia

1 %

2 %

Financing and other business activities, Asia

1 %

4 %

Financing and other business activities, France

9 %

2 %

Plastics and rubber manufacture, Asia

<1 %

2 %

Trade, France

23 %

5 %

Trade, India

<1 %

6 %

Trade, rest of Asia

1 %

2 %

Transport, Asia

2 %

8 %

Transport, France

3 %

1 %

Remaining activities

33 %

26 %

Total

100 %

100 %

3 Streamlined assessment of nonproduction-specific impacts

In this section, we shall show that although the number of social and economic impact pathways is large, it is possible to streamline the impact assessment when several impact pathways have the same endpoint, and the total impact on this endpoint can be measured independently from the contribution from each impact pathway. In this situation, detailed inventory and impact pathway descriptions may turn out to be unnecessary. In practice, this situation is encountered for a group of impacts that have very high overall importance, namely those that can be classified as nonproduction-specific impacts, i.e., social and economic impacts that are purely location specific and unrelated to enterprise-specific actions and choice of technology, which is in stark contrast to most biophysical impacts.

3.1 Income redistribution

One such impact is the income redistribution impact. Most productive activities imply a transfer of income between societal groups. The income levels of the donor and recipient groups are typically well known or easy to estimate and the distributional impact is then calculated as the increase (or loss) in utility caused by the transfer, by weighting the spending and income for each group by their relative marginal utility of income:
$$ Utility=\left(\frac{averageIncome}{subgroupIncome}\right)\hat{\mkern6mu} \delta $$

where both incomes are corrected for purchase power and δ is the elasticity of marginal utility of income (Lambert 2001). For our example calculations we have used a value for δ of 1.24, with a 95 % confidence interval of 1.14–1.35, based on Layard et al. (2008, p. 1856), who calculate this value from six surveys that relate well-being to income.

We illustrate the results for the clothing life cycle example in Table 2. Only the income effect of the payment from consumers to workers is included. Secondary consequences (multiplier effects) beyond the transfer between consumers and workers are not considered here. The values for the average and subgroup incomes were calculated from the industry- and country-specific data in EXIOBASE v.1 (Tukker et al. 2013) on work hours and value added, specified in worker’s compensation at three skill-levels, taxes and operating surplus. Taxes and operating surplus were added proportionally to the worker’s compensation for each skill level to obtain a separate value added (=subgroup income2) per skill level of workers for each industry in each country. These subgroup incomes were corrected for purchase power using the World Bank’s ratios of purchase power parity to the EUR market exchange rate. The utility for each separate subgroup was then calculated using the above formula, and averaged for each industry in each country. The data in Table 2 are then further aggregated across skill levels and some industries and countries for ease of presentation. The variation in distributional impact is much larger between countries than between industries and skill levels within one country.
Table 2

Redistribution impact of a French consumer’s expenditure of 1  EUR on an average clothing item in year 2000. Data calculated from detailed country- and industry-specific wage data from Exiobase v.1 (Tukker et al. 2013) and using an elasticity of income of 1.24, a global average purchase-power-corrected income of 8.57  EUR/work-hour, and 38  EUR/work-hour for the French consumer. Values are rounded

Activity

Value added (EUR2000)

Utility-weighted value (EUR2000,PPP) ( = redistribution impact)

Proportion between utility-weighted and original value

Animal products, China

0.002

0.017

8.6

Apparel and textile manufacture, rest of Europe

0.071

0.030

0.4

Apparel and textile manufacture, France

0.134

0.025

0.2

Apparel and textile manufacture, India

0.007

0.052

7.4

Apparel and textile manufacture, rest of Asia

0.062

0.056

0.9

Chemicals manufacture, Asia

0.005

0.022

4.4

Cultivation of plant-based fibres, India

0.0002

0.001

9.1

Electricity, Asia

0.004

0.012

2.8

Extraction of crude petroleum, Asia

0.005

0.011

2.0

Financing and other business activities, Asia

0.007

0.023

3.2

Financing and other business activities, France

0.089

0.017

0.2

Plastics and rubber manufacture, Asia

0.003

0.013

3.7

Trade, France

0.227

0.040

0.2

Trade, India

0.002

0.031

17.3

Trade, rest of Asia

0.006

0.013

2.0

Transport, Asia

0.017

0.046

2.7

Transport, France

0.029

0.005

0.2

Remaining activities

0.328

0.153

0.5

Total income

1.000

0.569

0.6

Total expenditure by French consumer

1.000

0.158

0.2

In a supply chain, the transfer is between final consumers of the product and the workers that produce the value added product. The payment by the final consumer is matched by the utility of the product for the consumer. For the workers, the payment received is for the utility of the work delivered. The income redistribution impact occurs because of the difference between the unweighted value added in the first column of Table 2 and the utility-weighted values in the second column. The total redistribution impact in Table 2 is 0.411 EUR, namely the sum of the utility-weighted income minus the utility-weighted expenditure of the consumer. If there had been no income inequalities, the utility-weighted values would have been the same as in the value added column of Table 2, i.e., the income and the expenditure would have been of equal size and there would have been no redistribution impact. While the utility-weighted income is 0.569 on average, this is composed of very high values for the income received in Asia (where the wages is generally much lower than the global average but with a high variability between countries and industries) and low values for the spending and the income received from the activities in Europe (where the wages are above the global average).

The distributional impact is typically positive (i.e., beneficial) for export-oriented activities of low-income countries, as shown by the example of the clothing life cycle in Table 2, where the total utility-weighted income is higher than the total expenditure by the French consumer. Utility is here used as a summary concept that captures the essential aspects of well-being. The modelled change in utility relates directly to the change in income, and is independent of what the money is spent for, i.e., irrespectively of whether it is spent on health care or luxury articles. Calculating the distributional impact directly as the change in utility can thus be seen as a top-down or summary approach to the impact pathway description, which circumvents the modelling of the specific relationships between income or income inequality and each specific aspect of utility, such as life expectancy (Feschet et al. 2013) and infant mortality (Bocoum et al. 2015), a modelling that is confounded by many unknown variables and intermediate steps.

3.2 Productivity impact of missing governance

The benefits from the above-described income redistribution impact are typically more than offset by the cost of other nonproduction-specific impacts, which can be grouped under the heading of “impacts of missing governance”. These impacts are either related directly to loss of productivity or to loss of well-being, or to both:
  • Direct productivity impacts include missing education, trade barriers, underemployment, corruption, and lacking physical infrastructure.

  • Well-being impacts that can be valued in terms of productivity include health impacts, lacking social infrastructure, and ecosystem and heritage impacts.

Obviously, a part of these impacts may be influenced directly by specific enterprises and/or technologies, but the point made here (see also the section “The relative importance of non-production-specific impacts”) is that the largest part of these impacts are not enterprise- or technology-specific. For example, an enterprise can directly contribute to education and health of their workers but cannot directly affect the general access to education and health care in a country that does not have an adequate governance structure for this. Probably the only impact category where direct influence of technology and enterprise is dominating is that of ecosystems impacts, which is generally valued at a few percent of the overall impacts on well-being (Weidema 2009).

When productivity impacts are internalised by the establishment of good governance, this leads either to higher wages (directly compensating the productivity impact) or to higher taxes (to pay for the currently missing public services), or both, and may also involve a net reduction in rent payments, while increasing productive efficiency. For example, removal of trade barriers directly removes the rent payments to the producer’s in the importing countries, directly increases wages in the exporting countries, and indirectly increases the value of wages in the importing countries through the lower prices, while increasing overall productive efficiency by removing inefficient producers in the importing country.

If all productivity impacts were internalised, the average expenditure per labor year would equal the potential value added per capita under good governance. Thus, at the aggregated level, the productivity impact of missing governance can be measured as the difference between the actual value added and the potential value added when all productivity impacts are internalised. We here assume that the average potential value added, i.e., the average potential productivity per work hour, is the same for all countries. We thus assume that the current productivity differences per work hour, except those that are caused by differences in inherent skills, are caused by differences in externalities.3

For the potential value added, we can thus use the same value for all countries, and for the clothing life cycle example in Table 3, we use a value of 1.87 times the current value added per work hour in the USA, taken from the calculation done in Weidema (2009) for the potential productivity, where the value 1.87 is calculated by correcting the national production for the current impacts on the US economy from unemployment and underemployment, health impacts, trade barriers, and missing education. Except for these impacts, the current difference between the USA and the global average is assumed to be due to lacking physical and social infrastructure. From EXIOBASE v.1 (Tukker et al. 2013), we have the value added per work hour for USA of 39.2 EUR2000 and the potential value added therefore becomes 1.87 * 39.2 EUR2000 = 73.3 EUR2000, which is 9.35 times the actual global average in year 2000. Without productivity impacts, we would thus expect a (counter-factual) global productivity in year 2000 valued at 9.35 times the actual.
Table 3

Productivity impact of missing governance related to a French consumer’s expenditure of 1  EUR on an average clothing item in year 2000. The productivity impact reflects the additional costs of the clothing item, if all productivity-reducing externalities were internalised. The redistribution impact from Table 2 has been included here for comparison. Overall, the productivity impact largely outweigh the redistribution impact

Activity

Productivity impact of missing governance (EUR2000,PPP)

Utility-weighted productivity impact (EUR2000,PPP)

Redistribution impact (from Table 2) (EUR2000,PPP)

Animal products, China

0.13

1.09

0.017

Apparel and textile manufacture, rest of Europe

0.39

0.29

0.030

Apparel and textile manufacture, France

0.12

0.023

0.025

Apparel and textile manufacture, India

1.11

8.22

0.052

Apparel and textile manufacture, rest of Asia

2.13

1.99

0.056

Chemicals manufacture, Asia

0.23

2.31

0.022

Cultivation of plant-based fibres, India

0.01

0.07

0.001

Electricity, Asia

0.23

0.51

0.012

Extraction of crude petroleum, Asia

0.18

0.33

0.011

Financing and other business activities, Asia

0.34

0.94

0.023

Financing and other business activities, France

0.08

0.016

0.017

Plastics and rubber manufacture, Asia

0.11

0.50

0.013

Trade, France

0.21

0.037

0.040

Trade, India

0.28

4.81

0.031

Trade, rest of Asia

0.21

0.51

0.013

Transport, Asia

0.71

3.00

0.046

Transport, France

0.03

0.005

0.005

Remaining activities

2.86

5.89

0.153

Sum for 1 EUR2000 expenditure on clothing

9.35

30.54

0.569

A country-specific factor for the productivity impact of missing governance is then calculated as the difference between the global potential value added per work hour and the actual national averages of value added (purchase-power-corrected) per work hour calculated from EXIOBASE v.1 (Tukker et al. 2013), and distributed over the industries of each country in proportion to their value added. The resulting values per industry are used for the column “productivity impact of missing governance” in Table 3.

In the ideal situation, where all productivity impacts are internalised and thereby either eliminated or compensated, income redistribution is not relevant. However, in the current situation, where productivity and income does vary significantly both between and within countries, the same marginal reduction in productivity impact or marginal increase in compensation would have larger utility in a poor than in a rich circumstance. Thus, both the impacts of missing governance and any compensating interventions should be utility-weighted with the same weights as applied for the income redistribution. Table 3 therefore also shows the utility-weighted productivity impact.

Table 3 shows particularly large productivity impacts in the Asian textile and apparel production. The utility-weighting give a particular emphasis on the impacts in India.

3.3 The relative importance of nonproduction-specific impacts

The productivity impact described above is a kind of “top-down summary indicator” for all the different impacts that cause loss of productivity, including missing education, health impacts, trade barriers, underemployment, and lacking physical and social infrastructure. While the summary indicator is all that is required for the calculation of the social footprint, it is necessary to have more details on the underlying impacts and their relative importance if prioritised action is to be taken to reduce the footprint. But also for this, a lot of data sources are readily available:

Nonproduction-specific well-being impacts can be measured as: “Number of people affected” × “Severity” × “Duration of impact”, and expressed in quality-adjusted life years (QALY), which may be monetarised.

For both productivity and well-being impacts, a co-responsibility exists for local enterprises because they benefit from the concurrent low internal costs of labour. Since these impacts are typically country-specific and unrelated to enterprise-specific actions and choice of technology, they can be measured at the national level and distributed over the enterprises in proportion to their value added.

To estimate the relative importance of the different impact categories, we recalculated the GDP multipliers from Weidema (2009) with more recent data and some slight changes in methodology as follows:
  • The impact of trade barriers has been estimated by Moavenzadeh (2013) to 2.5 and 5 % of GDP for the USA and the global economy, i.e., giving GDP multipliers of 1.025 and 1.05, respectively.

  • For unemployment, we have used the data from ILOSTAT for year 2013 (7.4 and 6 % unemployment rate for the USA and the world, respectively), assuming constant labour participation rates of 0.63 (labour force and population data from World Bank 2015), 3 % unavoidable frictional and structural unemployment, and full offset of employment hours in reduced household work reducing the increase in GDP by 18 and 21 % for the USA and the global economy, respectively, with relative value of unpaid work calculated from Ahmad and Koh (2011). The resulting GDP multipliers are 1 + 0.63*(7.4–3 %)*(1–18 %) = 1.023 for the USA and 1 + 0.63*(6–3 %)*(1–21 %) = 1.015 for the world average.

  • For health impacts, we have used the Global Burden of Disease (http://ghdx.healthdata.org/) health gaps for 2010 of 0.26 and 0.36 for the USA and the world, giving GDP multipliers of 1 + 0.26/(1–0.26) = 1.36 for the USA and 1 + 0.36/(1–0.36) = 1.56 for the world average.

  • The impact of missing education was calculated with the same methodology as in Weidema (2009), based on the work for the OECD by Psacharopoulos (1994), which has largely been confirmed by more recent research. This method estimates a 10 % increase in GDP per year of additional schooling until the 12th year, and a 6.8 % increase per year of additional schooling between the 12 and 18 years after which there is no further productivity effect. Based on the data from Barro and Lee (2010) for year 2010, the US and the world population aged 15 and above are estimated to have averages of 13.4 and 7.76 years of schooling, respectively, giving GDP multipliers for the potential of optimal schooling of (1.068)^(18–13.4) = 1.35 for the US and (1.1)^(12–7.76) + (1.068)^(18–12) = 3 for the world average.

The product of the above GDP multipliers for USA is 1.025 * 1.023 * 1.36 * 1.35 = 1.925 (compare to 1.87 in the calculation by Weidema 2009), which can be applied to the 2013 productivity of the USA of 62,370 USD 2013/capita, calculated from a GDP/capita of 53,040 USD2013 (Worldbank 2015) with a 17.6 % addition to account for the value of household production of according to the OECD household production database (Ahmad and Koh 2011). This brings the ideal productivity to 1.925 * 62,370 USD 2013/capita =120,000 USD2013.

The difference between the current (2013) global productivity of 12,860 USD 2013 (10,600 USD 2013 per capita GDP plus 21 % for the value of household production4) and the above ideal productivity of 120,000 USD2013/capita is explained partly by the product of the above global GDP multipliers (1.05 * 1.015 * 1.56 * 3 = 5), which together explain 74 % of the difference (56 % from education, 16 % from health, 1.5 % from trade barriers and 0.5 % from unemployment). The remaining impact can roughly be divided on missing physical infrastructure (7 %), missing social infrastructure (14 %) and ecosystem impacts (5 %), cf. Weidema (2009). Each of these impact categories can of course be further subdivided, using additional data sources as discussed in Weidema (2006a).

What is particularly noteworthy is that only ecosystem impacts and a share of the health impacts are technology- or enterprise-specific, while the remaining—at least 85 % of all—impacts are rather governance-related and can therefore be quantified from national statistics without need to access detailed technology- or enterprise-specific data.

4 The importance of inequality for valuing changes in quality-adjusted time

That well-being can be valued in terms of productivity rests on the budget constraint, i.e., that aggregate consumption expenditures cannot exceed the aggregate income from production, under steady state conditions of no net savings. Since aggregate consumption expresses the cost of all activities performed to ensure the current level of well-being, the budget constraint also provides an upper constraint on well-being. However, valuing well-being exclusively in terms of aggregate consumption and utility of income ignores the effect of inequality on the utility of consumption. Under inequality, a change in the amount of quality-adjusted time will have an impact on total utility that is independent from changes in income or consumption expenditure.

This is illustrated in Fig. 1, which shows how the income and expenditure are related to quality-adjusted time. The areas of the two boxes are the same, because the average level of consumption is dependent on income. If income from production is increased, here illustrated by a hypothetical 10 % increase, the consequences for the consumption can be one of two (and anything in between):
  1. A)

    The annual consumption is increased by 10 % and there is no change in quality-adjusted time (consumption box increases 10 % upwards on the Y-axis)

     
  2. B)

    The annual consumption stays the same and the amount of quality-adjusted time is increased by 10 % (consumption box increases 10 % along the X-axis). In practise, it may be a change in quality-adjusted time that induces the increase in income, rather than the income being spent on an increase in quality-adjusted time.

     
Fig. 1

Illustration of the different bases for calculation of the marginal utility of the same unweighted wellbeing when increasing consumption versus increasing quality-adjusted time (measured as kQ-hours or quality adjusted life years, QALYs)

In a circumstance of equality, where there is no utility-weighting of income and consumption, situation A and B would be assessed as equal in terms of well-being, because the unweighted areas have the same size. But in the (actual) circumstance of inequality, the two situations will be assessed differently, because of the difference in the marginal utility assigned to the two situations. In situation A, the marginal utility would be calculated relative to the current level of income, while in situation B, the calculation basis for the marginal increase would be between zero and the current level of income, i.e., on average half of the current level of income.

Using the example of France and India, two of the locations identified as hotspots in the above-mentioned life cycle but with a significant difference in income level, we show in Table 4 the effect of this difference on the marginal utility of income without and with a corresponding change in QALYs.
Table 4

The average marginal utility of income in France and India, year 2000, without change in amount of QALYs and when fully reflecting a change in the amount of QALYs. Calculated using a global average purchase -power-corrected value added of 8.57  EUR/work-hour (D), and an elasticity of marginal utility of income of 1.24 (E)

 

Value added; EUR2000,PPP per work hour (C)

Marginal utility of income (D/C)^E

France, only income change, on country average

38

0.16

India, only income change, on country average

1.8

7

France, only QALY change; on 0.5 * country average

19

0.37

India, only QALY change; on 0.5 * country average

0.9

16

The implication of this is that, under inequality, an intervention that changes the amount of QALYs for a population group will always give a larger change in (utility-weighted) well-being than an intervention of the same monetary value that only affects the level of consumption of the same population group. This would also imply that if the impact of missing governance (see the section “productivity impact of missing governance”) could be subdivided in direct productivity impacts and well-being impacts, then the latter should be utility-weighted with a larger value than that applied in Table 3, so that also the overall utility-weighting would become larger. Such a general subdivision would require more specific data and impact pathway modelling than what is available to us today, but could be performed for specific important stages in the life cycle of specific improvement options.

5 The potential credits for positive action

The potential for positive action is by far the largest in countries with missing governance. If an enterprise accepts the same overall expenditure in a poor country with missing governance as the enterprise accepts in a rich country with good governance, it is possible with this expenditure to change the productivity and well-being in the poor country much more than it will ever be possible in the rich country.

For the French clothing example, with its hotspots in India and the rest of Asia (see Table 1), with very high levels of utility of income in the Indian textile and apparel production and trade (see Table 2) and corresponding potentials for improvements in productivity (Table 3), especially when also affecting quality-adjusted life-years (Table 4), it would be obvious to point to some potentials for shared value improvements in the Indian textile and apparel production and trade:
  • The labour force is composed largely of unorganised women (D’Ambrogio 2014; Kane 2015; Kabeer and Mahmud 2004). Organising the labour force and providing better working conditions, hygiene, health care, prevention and compensation for accidents, may in some cases lead to improvements in productivity that are larger than the direct costs, especially when organised as a communal effort that can directly harvest the value created. Such initiatives may of course be enhanced if the product can furthermore be sold as “fair trade”, providing additional finance for such investments.

  • Indian girls have one of the lowest schooling levels in the World (Barro and Lee 2010). Child labour in the clothing industry is not uncommon (Brown 2012). The cost of schooling is low compared to the later increase in productivity, and makes it an interesting option to provide schooling, especially for girls, against a profitable share in their later income.

The specific valuation of such initiatives obviously require enterprise- and site-specific data, in contrast to the valuation of the country-specific impacts, but when it comes to specific positive actions, it is also much more straightforward to obtain the required data from the involved parties.

Turning the clothing life cycle net positive, i.e., entirely eliminating the productivity impact, is not a realistic ambition in the short term, since this would require a 10-fold increase in price of the product, cf. Table 3. Although this may be reduced if focussing specifically on improvements that reduce the impact on QALYs, cf. Table 4, it would still require a substantial price increase. A more realistic perspective would be simply to reduce the impact of the clothing life cycle compared to the average competing products.

6 Discussion and conclusions

The productivity impact presented in this paper is a top-down summary indicator that provides a monetary value for the sum of all productivity-reducing externalities. It is based on the assumption that the current productivity differences per work hour between countries is caused exclusively by differences in externalities (see also footnote 3). Being an overall summary measure, the productivity impact indicator does not provide any detail on the specific causal factors, such as missing education, health impacts, trade barriers, unemployment and other aspects of missing physical and social infrastructure. Additional data sources are therefore required to disaggregate the summary indicator according to these causal factors. However, at least 85 % of these impacts are governance-related and can therefore be quantified from national statistics without need to access detailed technology- or enterprise-specific data. This allows comprehensive assessments of social footprints of products at much lower efforts than seen so far.

The current large income inequalities and the large difference in marginal utility of different income groups make it imperative to utility-weight the impacts and improvement options in order to direct the efforts to where they have the largest utility. This also emphasises the importance of changes to the amount of quality-adjusted time (QALYs) over changes that only affects the level of consumption.

Locating activities in an area with missing governance has both costs and benefits, but the co-responsibility costs for missing governance typically outweigh the social benefits of redistributing income. This could be seen as an argument for not locating enterprises in countries with missing governance. However, this argument is more than outweighed by the potential for positive action, which is by far the largest in countries with missing governance, thus providing a compelling argument for placing activities in countries with missing governance, provided that it allows the enterprise to follow an active strategy to create shared value.

Footnotes

  1. 1.

    Work-place related impacts (e.g. health, labour rights violations, discrimination, missing pay and social security) can best be expressed in relation to working time, while impacts on the community (e.g. corruption, underemployment, missing tax and infrastructure payments) can best be expressed in relation to value added (Weidema 2006b).

  2. 2.

    Income includes both direct income as wages or entrepreneurial profit and indirect income via tax transfers, which together sum up to the value added.

  3. 3.

    It is an interesting question whether some of the difference in productivity should be ascribed to climate or other local conditions, also in a situation without externalities. Evidence presented by Dell et al. (2008) shows an effect of temperature on aggregate productivity for poor countries, but not for rich countries. Rather than a direct causal relationship between temperature and productivity, the adverse temperatures appear to aggravate other causes for low productivity. The question is therefore if such differences would persist in a situation without externalities (and therefore also without limitations on the movement of labour and with adequate income to provide temperature adaptation). If indeed they should persist, they would be related to variation in the work-leisure preference, i.e., the extent to which leisure is preferred to work or vice versa, and while such variation may affect the per capita GDP, it does not affect the productivity per work hour, which is what we are concerned with here. The same argument can be applied to any cultural differences in productivity that may remain after all externalities have been internalised.

  4. 4.

    The 21 % is calculated from a wage for household workers of 2.325 USD and 1312  h of unpaid work for each of the 73.7 % of the population above 15 years (Worldbank 2015).

Notes

Acknowledgments

We are grateful to Miguel Brandão for constructive sparring during the development of the arguments presented in this article and for nudging us to submit it to this special issue. The ideas were originally presented orally to the 2015 SETAC-Europe Annual meeting in Barcelona (Weidema 2015). We are also grateful for detailed and helpful comments from three anonymous reviewers of the originally submitted manuscript.

References

  1. Ahmad N, Koh S-H (2011) Incorporating estimates of household production of non-market services into international comparisons of material well-being. STD/2011/07. OECD, ParisGoogle Scholar
  2. D’Ambrogio E (2014) Workers’ conditions in the textile and clothing sector: just an Asian affair? European Parliament Briefing. Brussels: European Parliamentary Research Service. www.europarl.europa.eu/RegData/etudes/BRIE/2014/538222/EPRS_BRI(2014)538222_REV1_EN.pdf (Accessed 2015–07-30)
  3. Andrews ES, Barthel LP, Beck T, Benoît C, Ciroth A, Cucuzzella C, Gensch CO, Methot AL, Moberg A, Hébert J, Lesage P, Manhart A, Mazeau P, Mazjin B, Norris G, Parent J, Prakash S, Reveret JP, Spillemaeckers S, Ugaya S, Maria L, Valdivia S, Weidema BP (2009) Guidelines for social Life Cycle Assessment of Products - UNEP SETAC life cycle initiativeGoogle Scholar
  4. Barro R J, Lee J-W (2010) A New Data Set of Educational Attainment in the World, 1950–2010. NBER Working Paper No. 15902Google Scholar
  5. Benoit-Norris C, Cavan DA, Norris G (2012) Identifying social impacts in product supply chains: overview and application of the social hotspot database. Sustainability 4(9):1946–1965CrossRefGoogle Scholar
  6. Bocoum I, Macombe C, Revéret J-P (2015) Anticipating impacts on health based on changes in income inequality caused by life cycles. Int J Life Cycle Assess 20:405–417CrossRefGoogle Scholar
  7. Brown G (2012) Child labour & educational disadvantage – breaking the link, building opportunity. The Office of the UN Special Envoy for Global Education, LondonGoogle Scholar
  8. Ciroth A, Franze J (2011) LCA of an ecolabeled notebook - consideration of social and environmental impacts along the entire life cycle. GreenDeltaTC, BerlinGoogle Scholar
  9. Falque A, Feschet P, Garrabé M, Gillet C, Lagarde V, Loeillet D, Macombe C (2014) Social LCAs. Socio-economic effects in value chains. Montpellier: Fruitrop Thema, CIRAID. Copyright CIRAID 2013. English edition published 2014Google Scholar
  10. Feschet P, Macombe C, Garrabé M, Loeillet D, Saez A, Benhmad F (2013) Social impact assessment in LCA using the Preston pathway. Int J Life Cycle Assess 18:490–503CrossRefGoogle Scholar
  11. Grießhammer R, Buchert M, Gensch C-O, Hochfeld C, Manhart A, Reisch L, Rüdenauer I (2007) Prosa – Product Sustainability Assessment Guideline. (PROSA 2.0). Freiburg: öko-Institut e.v. – Institute for Applied EcologyGoogle Scholar
  12. Jørgensen A, Le Bocq A, Nazarkina L, Hauschild MZ (2008) Methodologies for social life cycle assessment. Int J Life Cycle Assess 13(2):96–103CrossRefGoogle Scholar
  13. Jørgensen A, Finkbeiner M, Jørgensen MS, Hauschild MZ (2010a) ) Defining the baseline in social life cycle assessment. Int J Life Cycle Assess 15(4):376–384CrossRefGoogle Scholar
  14. Jørgensen A, Lai L, Hauschild M (2010b) Assessing the validity of impact pathways for child labour and well-being in social life cycle assessment. Int J Life Cycle Assess 15:5–16CrossRefGoogle Scholar
  15. Kabeer N, Mahmud S (2004) Rags, riches and women workers: export-oriented garment manufacturing in Bangladesh. In: Carr M (ed) Chains of fortune: linking women producers and workers with global markets. Commonwealth Secretariat, London, pp. 133–164Google Scholar
  16. Kane G (2015) India Country Report. An Overview of the Garment Industry. Amsterdam: Clean Clothes Campaign. www.cleanclothes.org/resources/publications/factsheets/
  17. Klöpffer W (2008) Life cycle sustainability assessment of products. Int J Life Cycle Assess 13(2):89–95CrossRefGoogle Scholar
  18. Lambert, PJ (2001) The distribution and redistribution of income. Manchester University PressGoogle Scholar
  19. Layard R, Nickell S, Mayraz G (2008) The marginal utility of income. J Public Econ 92:1846–1857CrossRefGoogle Scholar
  20. Macombe C, Leskinen P, Feschet P, Antikainen R (2013) Social life cycle assessment of biodiesel production at three levels: a literature review and development needs. J Clean Prod 52:205–216CrossRefGoogle Scholar
  21. Moavenzadeh J, Doherty S, Philip R, Geiger T, Gottfredson M, Mattios G, Willink BT, Correa A, Hoekman B, Jackson S, Ferrantino M, Tsigas M (2013) Enabling trade. Valuing growth opportunities. World Economic Forum, GenevaGoogle Scholar
  22. Psacharopoulos G (1994) Returns to investment in education: a global update. World Dev 22(9):1325–1343CrossRefGoogle Scholar
  23. Reitinger C, Dumke M, Barosevcic M, Hillerbrand R (2011) A conceptual framework for impact assessment within SLCA. Int J Life Cycle Assess 16(4):380–388CrossRefGoogle Scholar
  24. Tukker A, de Koning A, Wood R, Hawkins T, Lutter S, Acosta J, Rueda Cantuche JM, Bouwmeester M, Oosterhaven J, Drosdowski T, Kuenen J (2013) EXIOPOL – development and illustrative analyses of a detailed global MR EE SUT/IOT. Econ Syst Res 25(1):50–70CrossRefGoogle Scholar
  25. Weidema BP (2006a) The integration of economic and social aspects in life cycle impact assessment. Int J Life Cycle Assess 11(1):89–96CrossRefGoogle Scholar
  26. Weidema, BP (2006b) Social impact categories, indicators, characterisation, and damage modelling. Presentation for the 29th Swiss LCA Discussion Forum 2006.07.15Google Scholar
  27. Weidema BP (2009) Using the budget constraint to monetarise impact assessment results. Ecol Econ 68(6):1591–1598CrossRefGoogle Scholar
  28. Weidema, BP (2015) The social footprint – A practical approach to comprehensive and consistent social LCA. Presentation to the SETAC 25th Annual meeting, Barcelona, 2015.05.3–7Google Scholar
  29. Worldbank (2015) Data. GDP (current LCU). GDP per capita (current US$). Labour force, total. Population, total. Population, ages 0–14 (% of total). Unemployment, total (% of total labor force). http://data.worldbank.org/indicator/NY.GDP.MKTP.CN/, NY.GDP.PCAP.CD/, SL.TLF.TOTL.IN/, SP.POP.TOTL/, SP.POP.0014.TO.ZS/, and SL.UEM.TOTL.ZS/ (accessed 2015–07-29)

Copyright information

© Springer-Verlag Berlin Heidelberg 2016

Authors and Affiliations

  1. 1.Danish Centre for Environmental AssessmentAalborg UniversityAalborgDenmark

Personalised recommendations